Case studiesManufacturing
Fewer defects on the line in manufacturing
How a Quebec potato grower and packer cut its optical sorter's error rate from 30 to 5 per cent within a few months, without replacing the machine.
- from 30%sorter error rate, from 30%
- 5%
- defect typeskinds of defect recognized
- 21
- efficiencygain in efficiency
- 25%+
- Industry
- Manufacturing
- Location
- Saint-Ubalde, Quebec
- AI partner
- Vooban
- Use case
- Sorting and grading
- The firm
- About 150 employees
What they did
The existing sorter stayed on the line. Vooban, an AI firm from Quebec City, fitted it with high-definition cameras and a deep-learning model that recognizes 21 kinds of defect, with a pipeline that retrains the model as the crop changes.
Why it matters to a business like yours
A sorter at 70 per cent meant staff re-sorting by hand and good product thrown out. At 95 per cent the line needs fewer hands, and the company sets the grade by customer, restaurant or grocer. The company reports a gain of at least 25 per cent in efficiency.
Relevant if
You grade, sort or inspect product by eye, or with a machine that still needs a person behind it.
“It makes it possible to eliminate routine tasks and provide more rewarding jobs.”
